Revise test commands in README.md

Updated test command syntax to use 'python -m unittest' for various test suites in README.

Author: bact

tests/README.md

@@ -50,7 +50,7 @@ The CI/CD test workflow is at
 
 ## Core tests (test_*.py)
 
-- Run `unittest tests.core`
+- Run `python -m unittest tests.core`
 - Focus on core functionalities.
 - Do not rely on external dependencies beyond the standard library.
 - Do not perform network access or corpus downloads.
@@ -59,22 +59,21 @@ The CI/CD test workflow is at
 
 ## Compact tests (testc_*.py)
 
-- Run `unittest tests.compact`
+- Run `python -m unittest tests.compact`
   - Need dependencies from `pip install "pythainlp[compact]"`
 - Test a limited set of functionalities that rely on a stable
   and small set of dependencies.
 - These dependencies are `PyYAML`, `nlpo3`, `numpy`, `pyicu`,
-  `python-crfsuite`, and `requests`.
-- Includes corpus download/remove tests (require network access and `requests`)
-  and any tests that depend on `numpy` or `python-crfsuite`.
+  and `python-crfsuite`.
+- Includes corpus download/remove tests (may require network access).
 - Tested on:
   - All OSes: earliest and second-latest supported Python versions
   - Ubuntu: additionally tested on the latest version
 - Test case class suffix: `TestCaseC`
 
 ## Extra tests (testx_*.py)
 
-- Run `unittest tests.extra`
+- Run `python -m unittest tests.extra`
   - Need dependencies from `pip install "pythainlp[compact,extra]"`
 - Test more functionalities that rely on larger set of dependencies
   or one that require more time or computation.
@@ -107,7 +106,7 @@ By separating tests by dependency group, we can:
 
 #### Umbrella suite: tests.noauto
 
-- Run `unittest tests.noauto`
+- Run `python -m unittest tests.noauto`
 - Includes all modular noauto test suites
 - Use this for comprehensive testing when all dependencies are available
 - Test case class suffix: `TestCaseN`
@@ -116,7 +115,7 @@ By separating tests by dependency group, we can:
 
 ##### PyTorch-based: tests.noauto_torch
 
-- Run `unittest tests.noauto_torch`
+- Run `python -m unittest tests.noauto_torch`
   - Need dependencies from `pip install "pythainlp[noauto-torch]"`
 - Tests requiring PyTorch and its ecosystem:
   - torch, transformers (PyTorch backend)
@@ -129,7 +128,7 @@ By separating tests by dependency group, we can:
 
 ##### TensorFlow-based: tests.noauto_tensorflow
 
-- Run `unittest tests.noauto_tensorflow`
+- Run `python -m unittest tests.noauto_tensorflow`
   - Need dependencies from `pip install "pythainlp[noauto-tensorflow]"`
 - Tests requiring TensorFlow:
   - deepcut tokenizer
@@ -139,7 +138,7 @@ By separating tests by dependency group, we can:
 
 ##### ONNX Runtime-based: tests.noauto_onnx
 
-- Run `unittest tests.noauto_onnx`
+- Run `python -m unittest tests.noauto_onnx`
   - Need dependencies from `pip install "pythainlp[noauto-onnx]"`
 - Tests requiring ONNX Runtime:
   - oskut, sefr_cut tokenizers
@@ -148,7 +147,7 @@ By separating tests by dependency group, we can:
 
 ##### Cython-compiled: tests.noauto_cython
 
-- Run `unittest tests.noauto_cython`
+- Run `python -m unittest tests.noauto_cython`
   - Need dependencies from `pip install "pythainlp[noauto-cython]"`
 - Tests requiring Cython-compiled packages:
   - phunspell spell checker
@@ -158,7 +157,7 @@ By separating tests by dependency group, we can:
 
 ##### Network-dependent: tests.noauto_network
 
-- Run `unittest tests.noauto_network`
+- Run `python -m unittest tests.noauto_network`
   - Need dependencies from `pip install "pythainlp[noauto-network]"`
 - Tests requiring network access:
   - Hugging Face Hub model downloads